Does ChatGPT recommend your brand? How to check, and how to fix it

How do you check whether ChatGPT recommends you?

Ask it what a shopper would ask. Not just your name, but the questions that come before a purchase. These are the five we used in our research.

1. What are the best brands for your category?The one that matters most. If you aren't named here, you aren't in the running.
2. What is your brand?Does it know you, and does it describe you correctly?
3. Where can I buy your brand?Your own website, a reseller, or nothing at all?
4. Is your brand legit?What it tells a shopper who is hesitating. Better you hear it first.
5. What does your brand sell?Whether it knows your range.

Use a temporary chat so your own history doesn't colour the answer. For each question, write down three things: were you named, was your own site given, and was what it said right. Be strict with the third. It's tempting to round up.

Why do the answers change from engine to engine?

Each engine has its own sources, its own search and its own view of which brands lead a category. In our AI shopping 2026 research, run on 26 October 2026, a brand that at least one assistant recommended was left out by 39% of the others, on average.

ChatGPT named a brand unprompted 40% of the time. Claude did so 60% of the time, Google AI 53%, DeepSeek and Meta AI 48% each, and Perplexity 44%. Checking one engine tells you about that engine and nothing else.

Why do the answers change from market to market?

An engine with web search looks for sources that fit the question, and a shopper in Germany asks in German, about prices in euros and where to buy locally. Different sources come back, so different brands get named. A brand the UK press covers well can be unknown to the same engine asked in Germany or the US.

The only way to know is to ask in each market you sell in, in that market's language. Translating your UK answers in your head doesn't count.

What if your brand name is an everyday word?

Then the engine may not think you're a brand at all. In our research, “What is Next?” got a dictionary definition from four of six assistants. “What is Iceland?” got the country from all six. “Dreams” got the PlayStation game or sleep from all six.

Funny, unless it's your brand. The fix is context. Say what you are in plain words wherever you're described, starting with your own site, and make sure your Organization data names your category.

What moves the answer?

Five fixes. The first four are on your own site, which is the good news.

1. Let the AI crawlers in.Check that robots.txt allows the crawlers the engines use for search and answers. A site that blocks them can't be read, however good it is. In our research, 21 of the 75 brand sites blocked our crawler outright. We doubt many of them meant to.
2. Add structured data.schema.org Organization on your homepage, Product and Offer on product pages. It tells an engine plainly who you are, what you sell and what it costs.
3. Publish llms.txt.A plain text file at the root of your site that says who you are and links your key pages, written for AI readers. Small, cheap and worth doing.
4. Keep your product data clean.Titles, prices, stock and variants that match your site, in a feed that updates. Shopping answers and agents read it directly. At most brands that feed is a mess nobody owns, so find it an owner.
5. Get written about.Engines learn about a brand from what other people publish about it: press, reviews, guides and lists. It's the slowest fix, and for a new brand it moves the answers most.

One more, for shopping agents: give them an address. Of the 50 sites in our research we could read, 15 published a UCP manifest and a storefront MCP endpoint, the standards agents use to find products and buy. Every one of the 15 runs on Shopify. If you're not on Shopify, Storefront gives you both from your product feed, live in minutes, for one flat monthly price and no cut of your sales.

How long does it take?

We ran our own score on humanforward.io on 25 October 2026 and got 35 out of 100. Not a number you put on a slide. Search scored 33 out of 40, AI answers 2 out of 60, and ChatGPT named us in 9% of the answers we checked. The basics on the site were in place. What was missing was other people writing about us.

That's the usual pattern. The site fixes take days. Being written about takes longer, and we don't know a shortcut that doesn't involve doing something worth writing about. We're publishing each new result as we go.

Can you check every engine at once?

Yes. The free score on our homepage shows where you stand in about a minute, with what the engines said and the fixes ranked. Findability Pro checks all eight engines, ChatGPT, Claude, Perplexity, Google AI (Gemini), Grok, DeepSeek, Meta AI (Muse) and Mistral Le Chat, in three markets every week. Start with the free one. If the answer is fine, you've lost a minute.